Rabbit R1: The First Personal AI AGENT Device NO ONE Saw Coming (Look Out, Apple)

Matthew Berman
10 Jan 202429:58

Summary

TLDRRabbit, a new startup, unveiled R1 - a dedicated AI assistant device powered by a proprietary 'large action model'. For $199 with no subscriptions, R1 features natural language interface, faster response times, and cross-platform app control to accomplish tasks. R1 can also learn new skills when taught. The founder contrasts R1's simplicity versus current app-driven phones, positioning it as a next generation companion versus a replacement.

Takeaways

  • 😲 A new company called Rabbit has launched an AI-powered device called the R1, which they claim is the future of computing
  • πŸ‘¨β€πŸ’» The R1 runs on Rabbit OS, which uses a Large Action Model (LAM) to understand natural language instructions and execute actions
  • πŸ’‘ The inspiration for the R1 came from analyzing how humans interact with apps and wanting to simplify the experience
  • πŸ€– The R1 can do things like play music, call rides, order food, book trips, and more through natural language instructions
  • 😎 The R1 has high-end hardware like a rotating camera, touchscreen, microphone, and global LTE
  • πŸ”¬ Rabbit created a 'Teach Mode' so users can show the R1 new skills that it doesn't already know
  • πŸ’° The R1 is competitively priced at $199 with no subscription, compared to other AI assistants
  • ⏱️ Rabbit claims the R1 responds over 10x faster than other voice AIs, within 500 milliseconds
  • πŸ‘ The presenter has already pre-ordered an R1 and is excited about the future of computing it represents
  • 😱 The presenter believes the R1 shows how apps may not be needed in the future as AI takes over more computing tasks

Q & A

  • What is the name of the new AI assistant device announced in the video?

    -The new AI assistant device announced in the video is called the Rabbit R1.

  • What is the price of the Rabbit R1 device?

    -The Rabbit R1 is priced at $199 with no subscription fees.

  • What is the main innovation of the Rabbit R1 compared to current smartphones?

    -The main innovation of the Rabbit R1 is that it uses a natural language interface and AI to accomplish tasks, instead of requiring users to navigate through apps.

  • What is the Lamb technology mentioned in the video?

    -Lamb stands for 'large action model'. It is a new type of AI foundation model created by Rabbit that understands human intentions on computers and can execute actions.

  • How does the Rabbit R1 protect user privacy?

    -The Rabbit R1 redirects users to login pages of third-party services instead of storing credentials. It does not create fake accounts or track user data.

  • What is the Teach Mode mentioned for the Rabbit R1?

    -The Teach Mode allows users to demonstrate new skills to the R1, so it can learn new capabilities beyond its initial training.

  • How does the Rabbit R1 compare in price to other AI devices?

    -At $199 the R1 is much cheaper than other AI devices like Anthropic's Claude ($699) and Alexa devices ($200+).

  • Why does the video suggest apps may not be needed in the future?

    -As AI like the R1's Lamb can learn to operate computers via natural language, apps may not be needed as an intermediary.

  • Who designed the hardware for the Rabbit R1 device?

    -The hardware design for the R1 was created in collaboration with Swedish firm Teenage Engineering.

  • When will the Rabbit R1 start shipping to customers?

    -Rabbit states the R1 will begin shipping to customers in Easter 2024.

Outlines

00:00

πŸ˜€ Introduction to Rabbit AI Assistant R1

The CEO of Rabbit, Jesse, introduces their new AI-powered device called R1. He explains how current smartphones with many apps are not intuitive anymore. R1 uses natural language and AI called Large Action Model to understand user intentions and accomplish tasks.

05:00

😊 How the Large Action Model Works

R1's Large Action Model or LAM allows it to learn and interact with any interface to accomplish tasks end-to-end for the user. This allows R1 to trigger actions across devices and platforms using natural language.

10:02

πŸ“± Demo of R1's Capabilities

Jesse demonstrates R1's abilities like playing music, ordering food, booking rides, planning trips, taking notes, translating languages, etc. R1 can understand complex requests and coordinate multiple apps to fulfill the user's needs.

15:03

πŸ‘οΈ Rabid Eye Computer Vision

R1 has a 360 degree camera called Rabid Eye which allows computer vision capabilities. It can analyze surroundings, read documents, and find objects in the refrigerator to assist the user.

20:03

πŸ€– Teach Mode for Custom Skills

R1 has an experimental Teach Mode where users can show R1 new skills step-by-step. This allows R1 to learn new abilities not originally programmed, expanding its knowledge.

25:04

πŸ’° R1 Priced at $199, Available Soon

R1 is priced at only $199 with no subscription, much cheaper than competitors. Preorders are open now for delivery by March 2024. Jesse is excited for people to experience R1.

Mindmap

The video is abnormal, and we are working hard to fix it.
Please replace the link and try again.

Keywords

πŸ’‘Rabbit R1

Rabbit R1 is the dedicated AI assistant device launched by the startup company Rabbit. It is marketed as the future of computing, with a revolutionary natural language interface and ability to take actions on the user's behalf across applications and platforms. The R1 demonstrates functions like playing music, booking rides, ordering food, and planning trips entirely through voice commands.

πŸ’‘large language model (LLM)

A large language model or LLM is a core AI technology that powers devices like the R1. LLMs have advanced natural language understanding that allows seamless human-computer interaction. The video suggests Rabbit's "large action model" is a type of LLM specialized for executing commands.

πŸ’‘natural language interface

The natural language interface on the R1 allows users to interact through voice and conversational commands instead of traditional app interfaces. This intuitive approach aims to simplify computing by removing the need to locate apps, navigate menus, or click buttons.

πŸ’‘end of programming

The host suggests the R1 represents a future where programming is no longer needed. With powerful AI that can parse commands and take end-to-end actions, manually coding software may be replaced by models that generate applications directly.

πŸ’‘teach mode

The R1 has an experimental "teach mode" where users can demonstrate workflows, like generating an image in an app. The device will then memorize and generalize these steps to complete similar tasks through natural language commands.

πŸ’‘form factor

The R1 introduces a new dedicated hardware form factor for AI assistants, contrasting smartphones that constrain assistants within existing app ecosystems. The unique design and specialization for natural interaction sets the R1 apart.

πŸ’‘computer vision

Computer vision capabilities on the R1 allow it to analyze visual data from its camera, like identifying objects in a refrigerator to suggest recipes. This increases the contexts and environments where the device can be helpful.

πŸ’‘privacy

Rabbit emphasizes privacy in the R1's design, avoiding concerns like fake user accounts. User data and credentials are kept isolated from Rabbit, retaining security and trust.

πŸ’‘future computing

The host positions the conversational R1 as the future of computing, where apps fade away in favor of AI that can operate systems entirely through natural language. This positions Rabbit as pioneering the next paradigm.

πŸ’‘specialization

Unlike smartphones that handle communication and computing, the R1 specializes in AI assistance. Rabbit believes focusing on natural interaction instead of an app ecosystem improves the experience.

Highlights

Rabbit R1 is a new AI-powered mobile device with no apps or interface, just voice

R1 uses a Large Action Model (LAM) to understand intentions and execute actions on computers

LAM learns to trigger actions across platforms like iOS, Android, desktop - a universal agent

R1 has faster response times than other voice AIs like Siri or Alexa

R1 can book rides, order food, and plan trips by coordinating across multiple apps

R1 has a teach mode where users can show it new skills

R1 translates conversations in real-time and takes visual actions via its camera

R1 is focused on being an AI companion, not replacing smartphones

R1 is priced at $199 with no subscription fees

R1 represents a new generation of AI-native devices

The end of programming is near as AI learns to operate computers directly

Apple and Google will likely build similar offerings, but constrained as smartphone features

Taking a new approach allows fundamental innovation unlike incremental changes

Major hardware companies will follow examples like R1 in rethinking products

Revolutionary products emerge when risk-taking founders completely reimagine computing

Transcripts

play00:00

a company called rabbit just launched

play00:02

what I believe is the future of

play00:04

computing it is an incredible device

play00:07

that is a dedicated AI assistant and

play00:11

with no rumors or press prior to

play00:13

yesterday the rabbit R1 has set the

play00:16

internet of Blaze take a look at this

play00:19

incredible demo because you are going to

play00:21

be blown

play00:22

[Music]

play00:24

away hi everyone my name is Jesse and

play00:27

I'm the founder and CEO of rabbit

play00:30

I'm so excited to be here today to

play00:33

present you two things we've been

play00:34

working on a revolutionary New

play00:37

Foundation model and a groundbreaking

play00:39

consumer mobile device powered by it our

play00:43

mission is to create the simplest

play00:46

computer something so intuitive that you

play00:49

don't need to learn how to use it the

play00:52

best way to achieve this is to break

play00:54

away from app-based operating system

play00:57

currently used by smartphones instead

play01:00

that we envision a natural language

play01:03

centered approach the computer we're

play01:06

building which we call a companion

play01:09

should be able to talk to understand and

play01:12

more importantly get things done for you

play01:15

the future of human machine interfaces

play01:18

should be more

play01:20

intuitive now before we get started

play01:23

let's take a look at the existing mobile

play01:25

devices that we use daily the one device

play01:28

that's in your pocket the smartphones

play01:31

like iPhone and Android

play01:34

phones these guys been here for years

play01:37

and we've grown tired of them the

play01:39

problem with these devices however is

play01:42

not the hardware phone factor it's

play01:44

what's inside the app based operating

play01:48

system want to get a right to the office

play01:51

there's the app for that want to buy

play01:53

groceries there's another app for that

play01:56

each time you want to do something you

play01:57

fumble through multiple pages and

play01:59

folders to find the app you want to use

play02:02

and there are always endless buttons

play02:04

that you need to click add to the cards

play02:07

go to the next page check the boxes and

play02:09

jumping back and forth and so on the

play02:12

smartphone was supposed to be intuitive

play02:14

but with hundreds of apps on your phone

play02:16

today that don't work together it no

play02:19

longer is if you look at the top ranking

play02:22

apps on App Stores today you'll find

play02:25

that most of them focus on entertainment

play02:28

our smartphones has become the best

play02:30

device to kill time instead of saving

play02:33

them it's just harder for them to do

play02:37

things many people before us have tried

play02:40

to build a simpler and more intuitive

play02:42

computers with AI a decade ago companies

play02:45

like apple Microsoft and Amazon made

play02:49

Siri contana and Alexa with these smart

play02:53

speakers often they either don't know

play02:56

what you're talking about or fail to

play02:58

accomplish the task

play03:00

we asked for recent achievements in

play03:03

large language models however or llms a

play03:07

type of AI technology have made it much

play03:10

easier for machines to understand you

play03:13

the popularity of L chatbox over the

play03:17

past years has shown that the natural

play03:19

language based experience is the PA

play03:23

forward however where this assistance

play03:26

struggle is still getting things done

play03:30

for example if you go to the chbt and

play03:34

use your Expedia plugin to book a ticket

play03:37

it can suggest options but ultimately

play03:40

cannot assist you in completing the

play03:42

booking process from start to finish

play03:45

things like Chad gbt are extremely good

play03:48

at understanding your intentions but

play03:51

could be better at trigging

play03:53

actions another Hot Topic is a field of

play03:56

research around what they call agents

play04:00

it has caught the eye of many open-

play04:02

Source projects and productivity

play04:05

software companies what remains to be

play04:07

solved is for these agents to perform

play04:10

tasks end to end accurately and speedily

play04:14

the problem is forcing a model to

play04:16

perform a task it is not designed for

play04:20

whether for a language model to reason

play04:22

about web page using super prompts or

play04:25

screenshots we have yet to produce an

play04:28

agent as good as users simply clicking

play04:31

the buttons to fulfill our vision of a

play04:35

delightful intuitive companion we must

play04:38

go beyond a piece of complex software we

play04:42

want it to be in the hands of

play04:45

everyone so we first set up to

play04:48

fundamentally understand how computer

play04:50

apps are structured and more importantly

play04:54

how humans interact with them we wanted

play04:57

to find a way for our AI to trigger

play05:00

auctions on behalf of users across all

play05:04

environments we want it to be Universal

play05:08

not just a chrome Plug-In or a limited

play05:10

set of apps but everything iOS Android

play05:14

and desktop these applications share

play05:17

something in common the interface they

play05:20

all have a user interface so at a

play05:24

philosophical level if we can make an AI

play05:27

trigger actions on any kind of interface

play05:30

just like a human would we will solve

play05:32

the problem this Insight led us to

play05:35

create the large action model or lamb as

play05:39

we call it it is new foundational model

play05:43

that understands and executes human

play05:45

intentions on computers driven by our

play05:48

research in neuros symbolic

play05:51

systems with a large action model we

play05:53

fundamentally find a solution to the

play05:56

challenges that apps apis or or agent

play06:00

phas we solve it with

play06:04

interfaces lamb can learn any interfaces

play06:07

from any software regardless of which

play06:10

platform they're running on in short the

play06:13

large language model understands what

play06:15

you say but a large action model get

play06:19

things done we use lamb to bring AI from

play06:22

words to

play06:24

action finally we can build a computer

play06:28

that in addition to understanding what

play06:29

you're trying to saying can actually

play06:32

help you do things on your

play06:34

behalf we pack the large action model

play06:37

into a advanced rabbit OS operating

play06:39

system it is built for realtime

play06:42

interactions between you and the rabbit

play06:45

powered by lamp large action models

play06:47

concept and test results are so powerful

play06:51

that we decided to make a one of kind

play06:54

mobile device introducing R1 your pocket

play06:58

companion

play06:59

[Music]

play07:33

[Music]

play07:48

[Music]

play07:58

this is the rabbit R1 designed in

play08:02

collaboration with teenage engineering

play08:04

the R1 is a fully Standalone device

play08:07

primarily driven by natural language the

play08:10

accumulation of countless hours of work

play08:13

from our research design and software

play08:16

teams it is simple to use R1 is equipped

play08:21

with everything you need to interact

play08:23

with you and perceive your surroundings

play08:26

it comes with a touchcreen

play08:29

a push to talk button and an analog

play08:32

scroll wheel a microphone and

play08:35

speakers and a computer vision enabled

play08:38

360 degree rotational camera we call the

play08:41

Rabid eye it is Bluetooth and Wi-Fi

play08:44

capable and I has the same car slots

play08:47

that supports Global 4G LTE network and

play08:51

now let's see how it works I can ask

play08:54

rabbit for anything just like if I'm

play08:56

using chbt the difference though is a

play08:59

speed with rabid OS I get a response 10

play09:03

times faster than most of the voice AI

play09:06

project rabbit answers my questions

play09:08

within 500 milliseconds with a push to

play09:12

talk button you don't need to say

play09:14

anything to wake it up just press and

play09:17

hold the button and talk like a

play09:19

walkie-talkie now let's give it a shot

play09:22

what's the nature of the

play09:24

reality the nature of reality is a topic

play09:26

that has captivated the minds of

play09:28

philosophers and iners throughout

play09:31

history according to Bertrand Russell

play09:33

reality is composed of hard data derived

play09:35

from sensory experiences and logical

play09:38

analysis it is the amalgamation of our

play09:40

perceptions thoughts and the external

play09:43

world I can speak or type with R1 if you

play09:46

shake R1 a keyboard will pop up you will

play09:50

also know that the response speat is

play09:53

much faster than using some of the L

play09:56

assistant apps with web browser such as

play09:58

ching

play10:00

what's the stock price of

play10:01

Coca-Cola searching for the stock price

play10:04

of

play10:06

Coca-Cola the stock price of Coca-Cola

play10:08

KO is

play10:11

$59.7 who played the role Oppenheimer in

play10:14

Christoper Nolan's latest movies in

play10:17

Christopher Nolan's most recent film

play10:19

Oppenheimer the role of Jay Robert

play10:21

Oppenheimer is played by Killian

play10:23

Murphy the film is a biographical

play10:25

Thriller that explores the story of

play10:27

American scientist Jay Robert himer and

play10:29

his involvement in the development of

play10:31

the atomic

play10:33

bomb now let's see how the large action

play10:36

model works with lamb rabbit R1 can

play10:40

interact with all kinds of applications

play10:43

introducing the rabbit hole wipe portal

play10:46

I can log into different services and

play10:48

unlock functionalities for R1 it's just

play10:51

like iCloud let's say I want to listen

play10:53

to some

play10:58

music

play11:00

with a rabbit hole web portal I have

play11:03

complete control over which Services I

play11:05

want to activate on my R1 and which

play11:07

provider I prefer for music I use

play11:11

Spotify so let's go ahead and connect my

play11:13

Spotify to the rabbit

play11:18

hole now that this is Spotify login page

play11:21

before I start the login process I want

play11:23

to emphasize our commitment to privacy

play11:26

the way we design the infrastructure for

play11:28

app that we interact with we don't

play11:31

create fake users or spam users we don't

play11:35

hack their infrastructures and of course

play11:38

we do not store any of your thirdparty

play11:41

credentials instead we redirect you to

play11:44

their login page we never save your

play11:47

username and password and we don't track

play11:50

your data the authentication happens on

play11:54

the respective

play11:55

apps now I'm logging with Spotify

play12:03

rabbit OS recognize that Spotify has

play12:06

being authenticated on my device before

play12:09

this presentation I

play12:12

pre-authenticated some other apps like

play12:14

uber and door

play12:20

Dash just choose your preferred services

play12:22

and loging through them to get

play12:24

authenticated on your device it's that

play12:26

easy now let's play some music

play12:29

play craftwork pocket

play12:32

calculator playing pocket calculator by

play12:35

craftwork for

play12:42

you play another song from the

play12:46

album sure playing computer love for

play12:52

you that's all the tracks from the Same

play12:55

album who wrote the lyrics for this song

play12:59

[Music]

play13:02

Ralph hutter and Emil schul wrote the

play13:03

lyrics for the song Computer Love by

play13:06

craftwork which band sampled this

play13:10

song it seems that Coldplay talk is the

play13:13

only known example of a song that

play13:14

sampled Computer Love what do you think

play13:17

about this song Computer Love by

play13:20

craftwork is widely regarded as a

play13:21

groundbreaking and influential song in

play13:23

the realm of electronic music its Fusion

play13:26

of catchy Melodies robotic vocals and

play13:28

inovative use of synthesizers has left a

play13:30

lasting impact on the genre I can also

play13:33

use R1 to call a ride get me ride from

play13:37

my office to home now of course I will

play13:39

book an Uber ride for you from your

play13:40

office to your

play13:43

home please confirm the

play13:46

ride I have six people with three

play13:48

blockages find me Uber that can fit all

play13:50

of us for six people and three pieces of

play13:53

luggage I recommend booking an Uber XL

play13:55

as it provides ample space for all

play13:57

passengers and luggage

play14:01

please confirm the ride the ride shows

play14:04

up I just hit confirm over's on my way I

play14:08

can also use R1 to order

play14:10

food get me a 12 in pizza from Pizza Hut

play14:14

deliver to here the most ordered option

play14:16

on the app is fine ordering a 12-in

play14:18

pizza from Pizza Hut since you mentioned

play14:21

that the most ordered option is fine I

play14:23

will select that for

play14:27

you I just created an order for a 12-in

play14:30

pizza it's going to be hand tossed with

play14:32

a classic marinara sauce and topped with

play14:34

regular cheese please confirm your

play14:38

order that sounds really good I just

play14:40

confirmed order

play14:42

here now I'm waiting for my pizza I can

play14:45

also make R1 to a set of more complex

play14:48

actions that may take multiple apps to

play14:51

finish on a smartphone for example if I

play14:54

want to plan a trip to London I can just

play14:56

describe what I wish in full to

play14:59

R1 I want to take my family to London

play15:02

it's going to be two of us and a child

play15:04

of age 12 we're thinking of January 30th

play15:07

to February 5th can you plan the entire

play15:10

trip for me we like cheap non-stop

play15:12

flights grouped seats a cool SUV and a

play15:16

nice hotel that has

play15:18

Wi-Fi exploring ticketing options to

play15:20

make your trip a

play15:27

reality

play15:33

for your trip I found various flight

play15:35

options a range of hotels to choose from

play15:37

and car rentals available please confirm

play15:40

each option individually for further

play15:41

details and booking so it's all been

play15:44

planned out I just confirm confirm

play15:46

confirm and that's it could you come up

play15:49

with a schedule for fun things to do

play15:51

while I'm over

play15:53

there on

play15:56

it I have prepared a detailed travel

play15:58

plan for your trip in summary you will

play16:01

be exploring London's iconic landmarks

play16:03

visiting museums enjoying delicious

play16:05

meals at local restaurants and

play16:07

experiencing the vibrant atmosphere of

play16:08

different

play16:11

neighborhoods navigation details are

play16:13

also prepared and I can help you book

play16:15

tickets for any events you'd like to

play16:17

attend R1 just planned the entire trip

play16:19

for me that's awesome but it seems like

play16:21

this is a little bit too intense can you

play16:23

plan a easy schedule for us sure I'm

play16:28

working on it

play16:38

please take a look at it and let me know

play16:40

what you

play16:42

think it gave me a more relaxed schedule

play16:45

every day with all the details sounds

play16:47

really good and I just confirmed

play16:51

that I can foresee a wonderful trip

play16:54

Isn't that cool huh R1 help me reach

play16:57

others just like a phone but smarter it

play17:01

come with a build-in real-time

play17:02

translator and a notaker removing the

play17:05

final barriers to communication R1

play17:08

automatically detects the language

play17:10

spoken in the environment and provide

play17:13

the bidirectional translation R1 has an

play17:16

i an onboard camera designed for

play17:19

advanced computer vision it can analyze

play17:21

surroundings and take actions in real

play17:24

time to activate the eye just double tap

play17:28

the button

play17:32

oh funny seeing you here

play17:38

Rick let me take a

play17:57

look

play18:06

fridge can you make me a nice dish

play18:08

that's low in

play18:10

calories let me

play18:18

see how about a green garden

play18:23

omelette it's a delicious and low

play18:25

calorie dish that combines the freshness

play18:27

of broccoli and cabbage with the

play18:28

creaminess of eggs here's a simple

play18:31

recipe for

play18:33

you nice it recognize all the stuff and

play18:36

gave me the actual recipes rabbit eye

play18:39

can also help you with your

play18:46

documents here's the table I've been

play18:48

working on for job questionnaire can you

play18:51

create an additional column that matches

play18:53

candidates who mentioned rabbit then

play18:55

their are questions about how they found

play18:57

us

play18:59

sure let me take a look at the table and

play19:01

add the matching column for

play19:09

you I've processed the table and sent

play19:12

you an email with the

play19:14

results okay now let's check our email I

play19:17

can continue to interact with rabid OS

play19:20

even Beyond R1 let's reply this email

play19:23

directory can you add another column

play19:29

that matches candidates who have

play19:33

included rabbit in their question and

play19:37

are La

play19:39

based I just reply the email h

play19:45

s and I got a refined version from rabid

play19:48

o through my

play19:51

email let's say I have a unique routine

play19:54

or task I cannot do on my phone R1 can

play19:57

do that too we are experimenting with

play20:00

what we call the teach mode just like

play20:03

how I can teach my friend how to

play20:05

skateboard I can show R1 how to do it

play20:07

and it will learn from me this means

play20:09

that any user regardless of technical

play20:12

background can teach R1 to learn new

play20:16

skills so you go to teach Mode start a

play20:19

new

play20:20

session today I will show you how to

play20:23

generate an image of puppy using me

play20:25

Journey from prompt using Discord first

play20:29

I will go to the servers page and click

play20:31

one my own servers since this is only a

play20:35

general image generation I'll go to Mid

play20:37

Journey text

play20:39

Channel then I will use the image

play20:43

command along with the

play20:46

prompt here I'm putting a cute baby wild

play20:52

dog with big eyes animated cartoon on

play20:56

rail 8K okay let's wait for a minute for

play21:00

the engine to start generating the

play21:19

images once it's done let's click on the

play21:22

image to get a

play21:27

link

play21:45

I will then explain to Rabbit OS how to

play21:48

use this rabbit and annotate it so that

play21:51

I can generate anything not just

play21:55

poppies so let's go back to our web

play21:57

portal

play21:58

submit request it takes seconds for the

play22:01

web portal to finish

play22:07

processing and that's it it's that

play22:09

simple now once we finish the training I

play22:12

can go back to my R1 now let's use mid

play22:15

Journey as I told you to generate a

play22:17

picture of a bunny in pixel art

play22:21

style certainly

play22:24

Jesse I will use mid journey to generate

play22:26

a picture of a bunny in pixel art style

play22:28

for you please give me a moment to

play22:31

create the

play22:44

image now here you go you got a image

play22:46

generated on M journey through teach

play22:50

mode watch learn and repeat that's teach

play22:53

mode it's that simple that's all the

play22:56

demos for today with land fast evolving

play23:00

my R1 will eventually help me to do

play23:02

things that can never be achieved on an

play23:05

app based phone speaking of the current

play23:08

APP based phones the first question we

play23:10

ask about ourselves is why would I need

play23:14

a new device if I already have a, iPhone

play23:17

my iPhone can't do any of this at all we

play23:21

do not build rabbit R1 to replace your

play23:24

phone it's just a different generation

play23:27

of devices the app based system were

play23:30

introduced more than 15 years ago and a

play23:32

new generation of native AI power

play23:35

devices are just getting started here's

play23:38

a quick recap R1 is our companion that

play23:41

hosts the large action model with

play23:44

natural language I can use it for a wide

play23:47

range of tasks ask anything direct

play23:51

actions complex actions AI enhanced

play23:54

video calls notaker translator with a

play23:58

rabid eye computer vision and

play24:01

experimental teach mode on the hardware

play24:05

perspective we got a 360 rotational

play24:09

camera a global 4G LTE SIM card a push

play24:14

the talk button and an analog scroll

play24:16

wheel one last thing what about the

play24:20

price now before we reveal our price I

play24:23

want to do a quick

play24:25

comparison here are some of the best

play24:27

phones on market right now you got

play24:29

iPhone you got latest version of Android

play24:32

phones we're looking at somewhere around

play24:34

700 to $1,000 for a top phone with an

play24:37

app based system I bought my new iPhone

play24:40

15 Pro Max last year and it's the same

play24:43

experience as my previous ones here are

play24:46

not so smart smart speakers they're

play24:48

asking roughly around

play24:51

$200 but they're all outdated and

play24:54

finally here are a couple of the new

play24:56

things with only large language models

play24:58

you got AI paying asking for

play25:01

$699 plus monthly subscriptions for

play25:03

their base models you got tab asking for

play25:07

$600 and you got meta reband glasses

play25:11

asking for roughly $300 remember these

play25:14

are the things with only large language

play25:16

model we still think these were too

play25:19

expensive we priced the rabbit R1 at

play25:24

$199 no subscription no hidden fees you

play25:29

can order the R1 now at

play25:32

rabbit. and we are shipping Easter

play25:37

2024 I can't wait for you to experience

play25:40

the R1 for

play25:42

yourself thank

play25:45

[Music]

play25:51

you now how cool is that I've already

play25:54

ordered one it's

play25:55

$199 with no sub subscription and that's

play25:59

an incredible price point as compared to

play26:01

the Humane AI pin which is I think $7 or

play26:04

$800 and you have to have a subscription

play26:07

for it not only that and I know this

play26:09

might be controversial I think it's

play26:11

gorgeous I love the color I love the

play26:13

design and it's by a design firm called

play26:15

teenage engineering which is known for

play26:17

their absolutely exceptional design now

play26:20

what's truly unique about this device is

play26:22

it's completely AI powered and I already

play26:25

ordered one and they're saying that

play26:27

it'll be deliver Ed at the end of March

play26:28

so that's really soon and this is fully

play26:31

AI powered and not only that it can

play26:33

actually learn actions so not only are

play26:35

you operating a computer using natural

play26:37

language but if you come across things

play26:40

that it doesn't know how to do you can

play26:41

teach it to do those tasks and I talked

play26:43

a lot about the end of programming and

play26:45

you're probably thinking what does that

play26:46

have to do with this well this is a

play26:48

device that really gives you a preview

play26:49

of what the future of computing looks

play26:52

like it is natural language directly to

play26:55

a large language model directly

play26:57

computing the end result or whatever the

play26:59

task you want to complete is it's going

play27:01

to do that for you no apps maybe not

play27:03

even an interface eventually just voice

play27:05

and execution now they don't call their

play27:07

AI an llm they call it an L which is a

play27:11

large action model now I don't know if

play27:13

that's just a marketing term for a large

play27:15

language model and they're just calling

play27:16

it something else but what it says here

play27:18

is lamb is a new type of foundation

play27:19

model that understands human intentions

play27:21

on computers and that sounds exactly

play27:24

like a large language model except maybe

play27:26

it's really good at function calling but

play27:27

again that's just a large language model

play27:29

and so they actually put together a

play27:31

bunch of their research it's not quite a

play27:33

white paper but they do talk about a lot

play27:35

of what they found and how they created

play27:36

this new device now I want to show you

play27:38

this real quick these are videos

play27:40

obviously a lot of them of people going

play27:42

through apps and everything from Google

play27:45

Maps to ordering food to Airbnb and this

play27:48

is how they trained the model to

play27:50

actually get things done and the model

play27:52

that they chose in the device itself is

play27:53

supposed to be broadly applicable it can

play27:56

accomplish pretty much any task so it

play27:58

was trained on a bunch of usage of

play28:00

different apps and then it abstracts

play28:02

that away with a large language model

play28:04

and then simplifies the interface to

play28:06

just be voice and you can see here kind

play28:08

of like self-operating computer it's

play28:10

actually mapping all the different

play28:11

buttons within all of the different apps

play28:14

it records it and then it tries to learn

play28:16

how to actually get to let's say if you

play28:18

want to play an Ed Sheeran song how to

play28:19

do that from within the Spotify app and

play28:22

again bringing it back to the end of

play28:24

programming this is kind of the

play28:26

intermediate step where the AI is

play28:27

learning how to operate apps itself

play28:30

without you actually needing to go and

play28:32

navigate through these apps but the next

play28:34

step is probably to remove the apps

play28:36

altogether then you have the large

play28:38

language model Computing directly and

play28:40

when that happens there are no more apps

play28:42

and then who is going to be building

play28:44

them and trust me I don't want this to

play28:46

happen but I can't foresee any other

play28:48

future now I'm sure a lot of you are

play28:49

saying hey couldn't this just be an app

play28:51

on the iPhone or an Android device and

play28:53

the answer is probably yes but they

play28:56

would have had to navigate at the many

play28:58

restrictions of being just an app in an

play29:01

app ecosystem on the iPhone on an

play29:03

Android device and then you're probably

play29:05

also saying hey isn't Apple or Android

play29:07

also going to launch something like this

play29:09

and again the answer is probably yes but

play29:11

they're going to be launching it as part

play29:13

of the iPhone or as part of an Android

play29:15

device and so they're shoving this new

play29:17

type of Technology this new interface

play29:19

into an existing product and I'm never

play29:22

as big of a believer in that as I am of

play29:24

taking a completely new approach to

play29:26

something and even even if rabbit R1

play29:28

itself does not end up taking off the

play29:30

form factor and how they're thinking

play29:32

about things is going to change how

play29:34

major consumer electronics companies

play29:37

think about building their own Hardware

play29:39

in the future and I always applaud

play29:41

Founders and companies that are willing

play29:42

to take huge risk and think about things

play29:44

in completely new ways because every

play29:46

once in a while we get something like

play29:48

the iPhone or we get something like the

play29:50

PC so this might be that if you enjoyed

play29:54

this video please consider giving a like

play29:55

And subscribe and I'll see you in the

play29:57

next one one